Merge branch 'main' into py_MinmaxStreamTTS_test

main/xiaozhi-server/core/connection.py

@@ -140,10 +140,6 @@ class ConnectionHandler:
         self.func_handler = None
 
         self.cmd_exit = self.config["exit_commands"]
-        self.max_cmd_length = 0
-        for cmd in self.cmd_exit:
-            if len(cmd) > self.max_cmd_length:
-                self.max_cmd_length = len(cmd)
 
         # 是否在聊天结束后关闭连接
         self.close_after_chat = False

main/xiaozhi-server/core/handle/helloHandle.py

@@ -1,5 +1,13 @@
+import time
 import json
+import random
 import asyncio
+from core.utils.dialogue import Message
+from core.utils.util import audio_to_data
+from core.providers.tts.dto.dto import SentenceType
+from core.utils.wakeup_word import WakeupWordsConfig
+from core.handle.sendAudioHandle import sendAudioMessage, send_stt_message
+from core.utils.util import remove_punctuation_and_length, opus_datas_to_wav_bytes
 from core.providers.tools.device_mcp import (
     MCPClient,
     send_mcp_initialize_message,
@@ -8,6 +16,17 @@ from core.providers.tools.device_mcp import (
 
 TAG = __name__
 
+WAKEUP_CONFIG = {
+    "refresh_time": 5,
+    "words": ["你好", "你好啊", "嘿，你好", "嗨"],
+}
+
+# 创建全局的唤醒词配置管理器
+wakeup_words_config = WakeupWordsConfig()
+
+# 用于防止并发调用wakeupWordsResponse的锁
+_wakeup_response_lock = asyncio.Lock()
+
 
 async def handleHelloMessage(conn, msg_json):
     """处理hello消息"""
@@ -30,3 +49,103 @@ async def handleHelloMessage(conn, msg_json):
             asyncio.create_task(send_mcp_tools_list_request(conn))
 
     await conn.websocket.send(json.dumps(conn.welcome_msg))
+
+
+async def checkWakeupWords(conn, text):
+    enable_wakeup_words_response_cache = conn.config[
+        "enable_wakeup_words_response_cache"
+    ]
+
+    # 等待tts初始化，最多等待3秒
+    start_time = time.time()
+    while time.time() - start_time < 3:
+        if conn.tts:
+            break
+        await asyncio.sleep(0.1)
+    else:
+        return False
+
+    if not enable_wakeup_words_response_cache:
+        return False
+
+    _, filtered_text = remove_punctuation_and_length(text)
+    if filtered_text not in conn.config.get("wakeup_words"):
+        return False
+
+    conn.just_woken_up = True
+    await send_stt_message(conn, text)
+
+    # 获取当前音色
+    voice = getattr(conn.tts, "voice", "default")
+    if not voice:
+        voice = "default"
+
+    # 获取唤醒词回复配置
+    response = wakeup_words_config.get_wakeup_response(voice)
+    if not response or not response.get("file_path"):
+        response = {
+            "voice": "default",
+            "file_path": "config/assets/wakeup_words.wav",
+            "time": 0,
+            "text": "哈啰啊，我是小智啦，声音好听的台湾女孩一枚，超开心认识你耶，最近在忙啥，别忘了给我来点有趣的料哦，我超爱听八卦的啦",
+        }
+
+    # 获取音频数据
+    opus_packets = audio_to_data(response.get("file_path"))
+    # 播放唤醒词回复
+    conn.client_abort = False
+
+    conn.logger.bind(tag=TAG).info(f"播放唤醒词回复: {response.get('text')}")
+    await sendAudioMessage(conn, SentenceType.FIRST, opus_packets, response.get("text"))
+    await sendAudioMessage(conn, SentenceType.LAST, [], None)
+
+    # 补充对话
+    conn.dialogue.put(Message(role="assistant", content=response.get("text")))
+
+    # 检查是否需要更新唤醒词回复
+    if time.time() - response.get("time", 0) > WAKEUP_CONFIG["refresh_time"]:
+        if not _wakeup_response_lock.locked():
+            asyncio.create_task(wakeupWordsResponse(conn))
+    return True
+
+
+async def wakeupWordsResponse(conn):
+    if not conn.tts or not conn.llm or not conn.llm.response_no_stream:
+        return
+
+    try:
+        # 尝试获取锁，如果获取不到就返回
+        if not await _wakeup_response_lock.acquire():
+            return
+
+        # 生成唤醒词回复
+        wakeup_word = random.choice(WAKEUP_CONFIG["words"])
+        question = (
+            "此刻用户正在和你说```"
+            + wakeup_word
+            + "```。\n请你根据以上用户的内容进行20-30字回复。要符合系统设置的角色情感和态度，不要像机器人一样说话。\n"
+            + "请勿对这条内容本身进行任何解释和回应，请勿返回表情符号，仅返回对用户的内容的回复。"
+        )
+
+        result = conn.llm.response_no_stream(conn.config["prompt"], question)
+        if not result or len(result) == 0:
+            return
+
+        # 生成TTS音频
+        tts_result = await asyncio.to_thread(conn.tts.to_tts, result)
+        if not tts_result:
+            return
+
+        # 获取当前音色
+        voice = getattr(conn.tts, "voice", "default")
+
+        wav_bytes = opus_datas_to_wav_bytes(tts_result, sample_rate=16000)
+        file_path = wakeup_words_config.generate_file_path(voice)
+        with open(file_path, "wb") as f:
+            f.write(wav_bytes)
+        # 更新配置
+        wakeup_words_config.update_wakeup_response(voice, file_path, result)
+    finally:
+        # 确保在任何情况下都释放锁
+        if _wakeup_response_lock.locked():
+            _wakeup_response_lock.release()

main/xiaozhi-server/core/handle/intentHandler.py

@@ -1,11 +1,12 @@
 import json
-import asyncio
 import uuid
+import asyncio
+from core.utils.dialogue import Message
+from core.providers.tts.dto.dto import ContentType
+from core.handle.helloHandle import checkWakeupWords
+from plugins_func.register import Action, ActionResponse
 from core.handle.sendAudioHandle import send_stt_message
 from core.utils.util import remove_punctuation_and_length
-from core.providers.tts.dto.dto import ContentType
-from core.utils.dialogue import Message
-from plugins_func.register import Action, ActionResponse
 from core.providers.tts.dto.dto import TTSMessageDTO, SentenceType
 
 TAG = __name__
@@ -23,10 +24,14 @@ async def handle_user_intent(conn, text):
         pass
 
     # 检查是否有明确的退出命令
-    filtered_text = remove_punctuation_and_length(text)[1]
+    _, filtered_text = remove_punctuation_and_length(text)
     if await check_direct_exit(conn, filtered_text):
         return True
 
+    # 检查是否是唤醒词
+    if await checkWakeupWords(conn, filtered_text):
+        return True
+
     if conn.intent_type == "function_call":
         # 使用支持function calling的聊天方法,不再进行意图分析
         return False

main/xiaozhi-server/core/handle/receiveAudioHandle.py

@@ -1,12 +1,11 @@
 import time
-import asyncio
 import json
-from core.handle.sendAudioHandle import send_stt_message
+import asyncio
+from core.utils.util import audio_to_data
+from core.handle.abortHandle import handleAbortMessage
 from core.handle.intentHandler import handle_user_intent
 from core.utils.output_counter import check_device_output_limit
-from core.handle.abortHandle import handleAbortMessage
-from core.handle.sendAudioHandle import SentenceType
-from core.utils.util import audio_to_data_stream
+from core.handle.sendAudioHandle import send_stt_message, SentenceType
 
 TAG = __name__
 
@@ -22,7 +21,6 @@ async def handleAudioMessage(conn, audio):
         if not hasattr(conn, "vad_resume_task") or conn.vad_resume_task.done():
             conn.vad_resume_task = asyncio.create_task(resume_vad_detection(conn))
         return
-
     if have_voice:
         if conn.client_is_speaking:
             await handleAbortMessage(conn)
@@ -31,13 +29,11 @@ async def handleAudioMessage(conn, audio):
     # 接收音频
     await conn.asr.receive_audio(conn, audio, have_voice)
 
-
 async def resume_vad_detection(conn):
     # 等待2秒后恢复VAD检测
     await asyncio.sleep(1)
     conn.just_woken_up = False
 
-
 async def startToChat(conn, text):
     # 检查输入是否是JSON格式（包含说话人信息）
     speaker_name = None
@@ -118,12 +114,13 @@ async def no_voice_close_connect(conn, have_voice):
 
 
 async def max_out_size(conn):
+    # 播放超出最大输出字数的提示
+    conn.client_abort = False
     text = "不好意思，我现在有点事情要忙，明天这个时候我们再聊，约好了哦！明天不见不散，拜拜！"
     await send_stt_message(conn, text)
     file_path = "config/assets/max_output_size.wav"
-    conn.tts.tts_audio_queue.put((SentenceType.FIRST, [], text))
-    play_audio_frames(conn, file_path)
-    conn.tts.tts_audio_queue.put((SentenceType.LAST, [], None))
+    opus_packets = audio_to_data(file_path)
+    conn.tts.tts_audio_queue.put((SentenceType.LAST, opus_packets, text))
     conn.close_after_chat = True
 
 
@@ -141,35 +138,25 @@ async def check_bind_device(conn):
 
         # 播放提示音
         music_path = "config/assets/bind_code.wav"
-        conn.tts.tts_audio_queue.put((SentenceType.FIRST, [], text))
-        play_audio_frames(conn, music_path)
+        opus_packets = audio_to_data(music_path)
+        conn.tts.tts_audio_queue.put((SentenceType.FIRST, opus_packets, text))
 
         # 逐个播放数字
         for i in range(6):  # 确保只播放6位数字
             try:
                 digit = conn.bind_code[i]
                 num_path = f"config/assets/bind_code/{digit}.wav"
-                play_audio_frames(conn, num_path)
+                num_packets = audio_to_data(num_path)
+                conn.tts.tts_audio_queue.put((SentenceType.MIDDLE, num_packets, None))
             except Exception as e:
                 conn.logger.bind(tag=TAG).error(f"播放数字音频失败: {e}")
                 continue
         conn.tts.tts_audio_queue.put((SentenceType.LAST, [], None))
     else:
+        # 播放未绑定提示
+        conn.client_abort = False
         text = f"没有找到该设备的版本信息，请正确配置 OTA地址，然后重新编译固件。"
         await send_stt_message(conn, text)
         music_path = "config/assets/bind_not_found.wav"
-        conn.tts.tts_audio_queue.put((SentenceType.FIRST, [], text))
-        play_audio_frames(conn, music_path)
-        conn.tts.tts_audio_queue.put((SentenceType.LAST, [], None))
-
-
-def play_audio_frames(conn, file_path):
-    """播放音频文件并处理发送帧数据"""
-    def handle_audio_frame(frame_data):
-        conn.tts.tts_audio_queue.put((SentenceType.MIDDLE, frame_data, None))
-
-    audio_to_data_stream(
-        file_path,
-        is_opus=True,
-        callback=handle_audio_frame
-    )
+        opus_packets = audio_to_data(music_path)
+        conn.tts.tts_audio_queue.put((SentenceType.LAST, opus_packets, text))

main/xiaozhi-server/core/handle/sendAudioHandle.py

@@ -1,6 +1,9 @@
 import json
-from core.providers.tts.dto.dto import SentenceType
+import time
+import asyncio
 from core.utils import textUtils
+from core.utils.util import audio_to_data
+from core.providers.tts.dto.dto import SentenceType
 
 TAG = __name__
 
@@ -28,13 +31,78 @@ async def sendAudioMessage(conn, sentenceType, audios, text):
 
 
 # 播放音频
-async def sendAudio(conn, audios):
-    if audios is None:
+async def sendAudio(conn, audios, frame_duration=60):
+    """
+    发送单个opus包，支持流控
+    Args:
+        conn: 连接对象
+        opus_packet: 单个opus数据包
+        pre_buffer: 快速发送音频
+        frame_duration: 帧时长（毫秒），匹配 Opus 编码
+    """
+    if audios is None or len(audios) == 0:
         return
-    # 如果audios不是opus数组，则不需要进行遍历，可以直接发送;这里需要进行流控管理，防止发送过快引发客户端溢出
+
     if isinstance(audios, bytes):
+        if conn.client_abort:
+            return
+
+        conn.last_activity_time = time.time() * 1000
+
+        # 获取或初始化流控状态
+        if not hasattr(conn, "audio_flow_control"):
+            conn.audio_flow_control = {
+                "last_send_time": 0,
+                "packet_count": 0,
+                "start_time": time.perf_counter(),
+            }
+
+        flow_control = conn.audio_flow_control
+        current_time = time.perf_counter()
+        # 计算预期发送时间
+        expected_time = flow_control["start_time"] + (
+            flow_control["packet_count"] * frame_duration / 1000
+        )
+        delay = expected_time - current_time
+        if delay > 0:
+            await asyncio.sleep(delay)
+
+        # 发送数据包
         await conn.websocket.send(audios)
 
+        # 更新流控状态
+        flow_control["packet_count"] += 1
+        flow_control["last_send_time"] = time.perf_counter()
+    else:
+        # 文件型音频走普通播放
+        start_time = time.perf_counter()
+        play_position = 0
+
+        # 执行预缓冲
+        pre_buffer_frames = min(3, len(audios))
+        for i in range(pre_buffer_frames):
+            await conn.websocket.send(audios[i])
+        remaining_audios = audios[pre_buffer_frames:]
+
+        # 播放剩余音频帧
+        for opus_packet in remaining_audios:
+            if conn.client_abort:
+                break
+
+            # 重置没有声音的状态
+            conn.last_activity_time = time.time() * 1000
+
+            # 计算预期发送时间
+            expected_time = start_time + (play_position / 1000)
+            current_time = time.perf_counter()
+            delay = expected_time - current_time
+            if delay > 0:
+                await asyncio.sleep(delay)
+
+            await conn.websocket.send(opus_packet)
+
+            play_position += frame_duration
+
 
 async def send_tts_message(conn, state, text=None):
     """发送 TTS 状态消息"""
@@ -52,7 +120,7 @@ async def send_tts_message(conn, state, text=None):
             stop_tts_notify_voice = conn.config.get(
                 "stop_tts_notify_voice", "config/assets/tts_notify.mp3"
             )
-            audios, _ = conn.tts.audio_to_opus_data(stop_tts_notify_voice)
+            audios = audio_to_data(stop_tts_notify_voice, is_opus=True)
             await sendAudio(conn, audios)
         # 清除服务端讲话状态
         conn.clearSpeakStatus()
@@ -72,7 +140,7 @@ async def send_stt_message(conn, text):
     display_text = text
     try:
         # 尝试解析JSON格式
-        if text.strip().startswith('{') and text.strip().endswith('}'):
+        if text.strip().startswith("{") and text.strip().endswith("}"):
             parsed_data = json.loads(text)
             if isinstance(parsed_data, dict) and "content" in parsed_data:
                 # 如果是包含说话人信息的JSON格式，只显示content部分

main/xiaozhi-server/core/handle/textHandle.py

@@ -1,154 +1,14 @@
-import json
-import time
-from core.handle.abortHandle import handleAbortMessage
-from core.handle.helloHandle import handleHelloMessage
-from core.providers.tools.device_mcp import handle_mcp_message
-from core.utils.util import remove_punctuation_and_length, filter_sensitive_info
-from core.handle.receiveAudioHandle import startToChat, handleAudioMessage
-from core.providers.tools.device_iot import handleIotDescriptors, handleIotStatus
-from core.handle.reportHandle import enqueue_asr_report
-import asyncio
+from core.handle.textMessageHandlerRegistry import TextMessageHandlerRegistry
+from core.handle.textMessageProcessor import TextMessageProcessor
 
 TAG = __name__
 
+# 全局处理器注册表
+message_registry = TextMessageHandlerRegistry()
+
+# 创建全局消息处理器实例
+message_processor = TextMessageProcessor(message_registry)
 
 async def handleTextMessage(conn, message):
     """处理文本消息"""
-    try:
-        msg_json = json.loads(message)
-        if isinstance(msg_json, int):
-            conn.logger.bind(tag=TAG).info(f"收到文本消息：{message}")
-            await conn.websocket.send(message)
-            return
-        if msg_json["type"] == "hello":
-            conn.logger.bind(tag=TAG).info(f"收到hello消息：{message}")
-            await handleHelloMessage(conn, msg_json)
-        elif msg_json["type"] == "abort":
-            conn.logger.bind(tag=TAG).info(f"收到abort消息：{message}")
-            await handleAbortMessage(conn)
-        elif msg_json["type"] == "listen":
-            conn.logger.bind(tag=TAG).info(f"收到listen消息：{message}")
-            if "mode" in msg_json:
-                conn.client_listen_mode = msg_json["mode"]
-                conn.logger.bind(tag=TAG).debug(
-                    f"客户端拾音模式：{conn.client_listen_mode}"
-                )
-            if msg_json["state"] == "start":
-                conn.client_have_voice = True
-                conn.client_voice_stop = False
-            elif msg_json["state"] == "stop":
-                conn.client_have_voice = True
-                conn.client_voice_stop = True
-                if len(conn.asr_audio) > 0:
-                    await handleAudioMessage(conn, b"")
-            elif msg_json["state"] == "detect":
-                conn.client_have_voice = False
-                conn.asr_audio.clear()
-                if "text" in msg_json:
-                    conn.last_activity_time = time.time() * 1000
-                    original_text = msg_json["text"]  # 保留原始文本
-                    filtered_len, filtered_text = remove_punctuation_and_length(
-                        original_text
-                    )
-                    # 识别是否是唤醒词
-                    is_wakeup_words = filtered_text in conn.config.get("wakeup_words")
-                    if not is_wakeup_words:
-                        # 上报纯文字数据（复用ASR上报功能，但不提供音频数据）
-                        enqueue_asr_report(conn, original_text, [])
-                        # 否则需要LLM对文字内容进行答复
-                        await startToChat(conn, original_text)
-        elif msg_json["type"] == "iot":
-            conn.logger.bind(tag=TAG).info(f"收到iot消息：{message}")
-            if "descriptors" in msg_json:
-                asyncio.create_task(handleIotDescriptors(conn, msg_json["descriptors"]))
-            if "states" in msg_json:
-                asyncio.create_task(handleIotStatus(conn, msg_json["states"]))
-        elif msg_json["type"] == "mcp":
-            conn.logger.bind(tag=TAG).info(f"收到mcp消息：{message[:100]}")
-            if "payload" in msg_json:
-                asyncio.create_task(
-                    handle_mcp_message(conn, conn.mcp_client, msg_json["payload"])
-                )
-        elif msg_json["type"] == "server":
-            # 记录日志时过滤敏感信息
-            conn.logger.bind(tag=TAG).info(
-                f"收到服务器消息：{filter_sensitive_info(msg_json)}"
-            )
-            # 如果配置是从API读取的，则需要验证secret
-            if not conn.read_config_from_api:
-                return
-            # 获取post请求的secret
-            post_secret = msg_json.get("content", {}).get("secret", "")
-            secret = conn.config["manager-api"].get("secret", "")
-            # 如果secret不匹配，则返回
-            if post_secret != secret:
-                await conn.websocket.send(
-                    json.dumps(
-                        {
-                            "type": "server",
-                            "status": "error",
-                            "message": "服务器密钥验证失败",
-                        }
-                    )
-                )
-                return
-            # 动态更新配置
-            if msg_json["action"] == "update_config":
-                try:
-                    # 更新WebSocketServer的配置
-                    if not conn.server:
-                        await conn.websocket.send(
-                            json.dumps(
-                                {
-                                    "type": "server",
-                                    "status": "error",
-                                    "message": "无法获取服务器实例",
-                                    "content": {"action": "update_config"},
-                                }
-                            )
-                        )
-                        return
-
-                    if not await conn.server.update_config():
-                        await conn.websocket.send(
-                            json.dumps(
-                                {
-                                    "type": "server",
-                                    "status": "error",
-                                    "message": "更新服务器配置失败",
-                                    "content": {"action": "update_config"},
-                                }
-                            )
-                        )
-                        return
-
-                    # 发送成功响应
-                    await conn.websocket.send(
-                        json.dumps(
-                            {
-                                "type": "server",
-                                "status": "success",
-                                "message": "配置更新成功",
-                                "content": {"action": "update_config"},
-                            }
-                        )
-                    )
-                except Exception as e:
-                    conn.logger.bind(tag=TAG).error(f"更新配置失败: {str(e)}")
-                    await conn.websocket.send(
-                        json.dumps(
-                            {
-                                "type": "server",
-                                "status": "error",
-                                "message": f"更新配置失败: {str(e)}",
-                                "content": {"action": "update_config"},
-                            }
-                        )
-                    )
-            # 重启服务器
-            elif msg_json["action"] == "restart":
-                await conn.handle_restart(msg_json)
-        else:
-            conn.logger.bind(tag=TAG).error(f"收到未知类型消息：{message}")
-    except json.JSONDecodeError:
-        await conn.websocket.send(message)
+    await message_processor.process_message(conn, message)

main/xiaozhi-server/core/handle/textHandler/abortMessageHandler.py

@@ -0,0 +1,16 @@
+from typing import Dict, Any
+
+from core.handle.abortHandle import handleAbortMessage
+from core.handle.textMessageHandler import TextMessageHandler
+from core.handle.textMessageType import TextMessageType
+
+
+class AbortTextMessageHandler(TextMessageHandler):
+    """Abort消息处理器"""
+
+    @property
+    def message_type(self) -> TextMessageType:
+        return TextMessageType.ABORT
+
+    async def handle(self, conn, msg_json: Dict[str, Any]) -> None:
+        await handleAbortMessage(conn)

main/xiaozhi-server/core/handle/textHandler/helloMessageHandler.py

@@ -0,0 +1,16 @@
+from typing import Dict, Any
+
+from core.handle.helloHandle import handleHelloMessage
+from core.handle.textMessageHandler import TextMessageHandler
+from core.handle.textMessageType import TextMessageType
+
+
+class HelloTextMessageHandler(TextMessageHandler):
+    """Hello消息处理器"""
+
+    @property
+    def message_type(self) -> TextMessageType:
+        return TextMessageType.HELLO
+
+    async def handle(self, conn, msg_json: Dict[str, Any]) -> None:
+        await handleHelloMessage(conn, msg_json)

main/xiaozhi-server/core/handle/textHandler/iotMessageHandler.py

@@ -0,0 +1,20 @@
+import asyncio
+from typing import Dict, Any
+
+from core.handle.textMessageHandler import TextMessageHandler
+from core.handle.textMessageType import TextMessageType
+from core.providers.tools.device_iot import handleIotStatus, handleIotDescriptors
+
+
+class IotTextMessageHandler(TextMessageHandler):
+    """IOT消息处理器"""
+
+    @property
+    def message_type(self) -> TextMessageType:
+        return TextMessageType.IOT
+
+    async def handle(self, conn, msg_json: Dict[str, Any]) -> None:
+        if "descriptors" in msg_json:
+            asyncio.create_task(handleIotDescriptors(conn, msg_json["descriptors"]))
+        if "states" in msg_json:
+            asyncio.create_task(handleIotStatus(conn, msg_json["states"]))

main/xiaozhi-server/core/handle/textHandler/listenMessageHandler.py

@@ -0,0 +1,63 @@
+import time
+from typing import Dict, Any
+
+from core.handle.receiveAudioHandle import handleAudioMessage, startToChat
+from core.handle.reportHandle import enqueue_asr_report
+from core.handle.sendAudioHandle import send_stt_message, send_tts_message
+from core.handle.textMessageHandler import TextMessageHandler
+from core.handle.textMessageType import TextMessageType
+from core.utils.util import remove_punctuation_and_length
+
+TAG = __name__
+
+class ListenTextMessageHandler(TextMessageHandler):
+    """Listen消息处理器"""
+
+    @property
+    def message_type(self) -> TextMessageType:
+        return TextMessageType.LISTEN
+
+    async def handle(self, conn, msg_json: Dict[str, Any]) -> None:
+        if "mode" in msg_json:
+            conn.client_listen_mode = msg_json["mode"]
+            conn.logger.bind(tag=TAG).debug(
+                f"客户端拾音模式：{conn.client_listen_mode}"
+            )
+        if msg_json["state"] == "start":
+            conn.client_have_voice = True
+            conn.client_voice_stop = False
+        elif msg_json["state"] == "stop":
+            conn.client_have_voice = True
+            conn.client_voice_stop = True
+            if len(conn.asr_audio) > 0:
+                await handleAudioMessage(conn, b"")
+        elif msg_json["state"] == "detect":
+            conn.client_have_voice = False
+            conn.asr_audio.clear()
+            if "text" in msg_json:
+                conn.last_activity_time = time.time() * 1000
+                original_text = msg_json["text"]  # 保留原始文本
+                filtered_len, filtered_text = remove_punctuation_and_length(
+                    original_text
+                )
+
+                # 识别是否是唤醒词
+                is_wakeup_words = filtered_text in conn.config.get("wakeup_words")
+                # 是否开启唤醒词回复
+                enable_greeting = conn.config.get("enable_greeting", True)
+
+                if is_wakeup_words and not enable_greeting:
+                    # 如果是唤醒词，且关闭了唤醒词回复，就不用回答
+                    await send_stt_message(conn, original_text)
+                    await send_tts_message(conn, "stop", None)
+                    conn.client_is_speaking = False
+                elif is_wakeup_words:
+                    conn.just_woken_up = True
+                    # 上报纯文字数据（复用ASR上报功能，但不提供音频数据）
+                    enqueue_asr_report(conn, "嘿，你好呀", [])
+                    await startToChat(conn, "嘿，你好呀")
+                else:
+                    # 上报纯文字数据（复用ASR上报功能，但不提供音频数据）
+                    enqueue_asr_report(conn, original_text, [])
+                    # 否则需要LLM对文字内容进行答复
+                    await startToChat(conn, original_text)

main/xiaozhi-server/core/handle/textHandler/mcpMessageHandler.py

@@ -0,0 +1,20 @@
+import asyncio
+from typing import Dict, Any
+
+from core.handle.textMessageHandler import TextMessageHandler
+from core.handle.textMessageType import TextMessageType
+from core.providers.tools.device_mcp import handle_mcp_message
+
+
+class McpTextMessageHandler(TextMessageHandler):
+    """MCP消息处理器"""
+
+    @property
+    def message_type(self) -> TextMessageType:
+        return TextMessageType.MCP
+
+    async def handle(self, conn, msg_json: Dict[str, Any]) -> None:
+        if "payload" in msg_json:
+            asyncio.create_task(
+                handle_mcp_message(conn, conn.mcp_client, msg_json["payload"])
+            )

main/xiaozhi-server/core/handle/textHandler/serverMessageHandler.py

@@ -0,0 +1,92 @@
+import asyncio
+import json
+from typing import Dict, Any
+
+from core.handle.textMessageHandler import TextMessageHandler
+from core.handle.textMessageType import TextMessageType
+from core.providers.tools.device_mcp import handle_mcp_message
+
+TAG = __name__
+
+class ServerTextMessageHandler(TextMessageHandler):
+    """MCP消息处理器"""
+
+    @property
+    def message_type(self) -> TextMessageType:
+        return TextMessageType.SERVER
+
+    async def handle(self, conn, msg_json: Dict[str, Any]) -> None:
+        # 如果配置是从API读取的，则需要验证secret
+        if not conn.read_config_from_api:
+            return
+        # 获取post请求的secret
+        post_secret = msg_json.get("content", {}).get("secret", "")
+        secret = conn.config["manager-api"].get("secret", "")
+        # 如果secret不匹配，则返回
+        if post_secret != secret:
+            await conn.websocket.send(
+                json.dumps(
+                    {
+                        "type": "server",
+                        "status": "error",
+                        "message": "服务器密钥验证失败",
+                    }
+                )
+            )
+            return
+        # 动态更新配置
+        if msg_json["action"] == "update_config":
+            try:
+                # 更新WebSocketServer的配置
+                if not conn.server:
+                    await conn.websocket.send(
+                        json.dumps(
+                            {
+                                "type": "server",
+                                "status": "error",
+                                "message": "无法获取服务器实例",
+                                "content": {"action": "update_config"},
+                            }
+                        )
+                    )
+                    return
+
+                if not await conn.server.update_config():
+                    await conn.websocket.send(
+                        json.dumps(
+                            {
+                                "type": "server",
+                                "status": "error",
+                                "message": "更新服务器配置失败",
+                                "content": {"action": "update_config"},
+                            }
+                        )
+                    )
+                    return
+
+                # 发送成功响应
+                await conn.websocket.send(
+                    json.dumps(
+                        {
+                            "type": "server",
+                            "status": "success",
+                            "message": "配置更新成功",
+                            "content": {"action": "update_config"},
+                        }
+                    )
+                )
+            except Exception as e:
+                conn.logger.bind(tag=TAG).error(f"更新配置失败: {str(e)}")
+                await conn.websocket.send(
+                    json.dumps(
+                        {
+                            "type": "server",
+                            "status": "error",
+                            "message": f"更新配置失败: {str(e)}",
+                            "content": {"action": "update_config"},
+                        }
+                    )
+                )
+        # 重启服务器
+        elif msg_json["action"] == "restart":
+            await conn.handle_restart(msg_json)

main/xiaozhi-server/core/handle/textMessageHandler.py

@@ -0,0 +1,21 @@
+from abc import abstractmethod, ABC
+from typing import Dict, Any
+
+from core.handle.textMessageType import TextMessageType
+
+TAG = __name__
+
+
+class TextMessageHandler(ABC):
+    """消息处理器抽象基类"""
+
+    @abstractmethod
+    async def handle(self, conn, msg_json: Dict[str, Any]) -> None:
+        """处理消息的抽象方法"""
+        pass
+
+    @property
+    @abstractmethod
+    def message_type(self) -> TextMessageType:
+        """返回处理的消息类型"""
+        pass

main/xiaozhi-server/core/handle/textMessageHandlerRegistry.py

@@ -0,0 +1,45 @@
+from typing import Dict, Optional
+
+from core.handle.textHandler.abortMessageHandler import AbortTextMessageHandler
+from core.handle.textHandler.helloMessageHandler import HelloTextMessageHandler
+from core.handle.textHandler.iotMessageHandler import IotTextMessageHandler
+from core.handle.textHandler.listenMessageHandler import ListenTextMessageHandler
+from core.handle.textHandler.mcpMessageHandler import McpTextMessageHandler
+from core.handle.textMessageHandler import TextMessageHandler
+from core.handle.textHandler.serverMessageHandler import ServerTextMessageHandler
+
+TAG = __name__
+
+
+class TextMessageHandlerRegistry:
+    """消息处理器注册表"""
+
+    def __init__(self):
+        self._handlers: Dict[str, TextMessageHandler] = {}
+        self._register_default_handlers()
+
+    def _register_default_handlers(self) -> None:
+        """注册默认的消息处理器"""
+        handlers = [
+            HelloTextMessageHandler(),
+            AbortTextMessageHandler(),
+            ListenTextMessageHandler(),
+            IotTextMessageHandler(),
+            McpTextMessageHandler(),
+            ServerTextMessageHandler(),
+        ]
+
+        for handler in handlers:
+            self.register_handler(handler)
+
+    def register_handler(self, handler: TextMessageHandler) -> None:
+        """注册消息处理器"""
+        self._handlers[handler.message_type.value] = handler
+
+    def get_handler(self, message_type: str) -> Optional[TextMessageHandler]:
+        """获取消息处理器"""
+        return self._handlers.get(message_type)
+
+    def get_supported_types(self) -> list:
+        """获取支持的消息类型"""
+        return list(self._handlers.keys())

main/xiaozhi-server/core/handle/textMessageProcessor.py

@@ -0,0 +1,41 @@
+import json
+
+from core.handle.textMessageHandlerRegistry import TextMessageHandlerRegistry
+
+TAG = __name__
+
+
+class TextMessageProcessor:
+    """消息处理器主类"""
+
+    def __init__(self, registry: TextMessageHandlerRegistry):
+        self.registry = registry
+
+    async def process_message(self, conn, message: str) -> None:
+        """处理消息的主入口"""
+        try:
+            # 解析JSON消息
+            msg_json = json.loads(message)
+
+            # 处理JSON消息
+            if isinstance(msg_json, dict):
+                message_type = msg_json.get("type")
+
+                # 记录日志
+                conn.logger.bind(tag=TAG).info(f"收到{message_type}消息：{message}")
+
+                # 获取并执行处理器
+                handler = self.registry.get_handler(message_type)
+                if handler:
+                    await handler.handle(conn, msg_json)
+                else:
+                    conn.logger.bind(tag=TAG).error(f"收到未知类型消息：{message}")
+            # 处理纯数字消息
+            elif isinstance(msg_json, int):
+                conn.logger.bind(tag=TAG).info(f"收到数字消息：{message}")
+                await conn.websocket.send(message)
+
+        except json.JSONDecodeError:
+            # 非JSON消息直接转发
+            conn.logger.bind(tag=TAG).error(f"解析到错误的消息：{message}")
+            await conn.websocket.send(message)

main/xiaozhi-server/core/handle/textMessageType.py

@@ -0,0 +1,11 @@
+from enum import Enum
+
+
+class TextMessageType(Enum):
+    """消息类型枚举"""
+    HELLO = "hello"
+    ABORT = "abort"
+    LISTEN = "listen"
+    IOT = "iot"
+    MCP = "mcp"
+    SERVER = "server"

main/xiaozhi-server/core/providers/asr/base.py

@@ -1,18 +1,18 @@
 import os
+import io
 import wave
 import uuid
+import json
+import time
 import queue
 import asyncio
 import traceback
 import threading
 import opuslib_next
-import json
-import io
-import time
 import concurrent.futures
 from abc import ABC, abstractmethod
 from config.logger import setup_logging
-from typing import Optional, Tuple, List, Dict, Any
+from typing import Optional, Tuple, List
 from core.handle.receiveAudioHandle import startToChat
 from core.handle.reportHandle import enqueue_asr_report
 from core.utils.util import remove_punctuation_and_length
@@ -87,11 +87,9 @@ class ASRProviderBase(ABC):
             
             # 预先准备WAV数据
             wav_data = None
-            # 使用连接的声纹识别提供者
             if conn.voiceprint_provider and combined_pcm_data:
                 wav_data = self._pcm_to_wav(combined_pcm_data)
             
-            
             # 定义ASR任务
             def run_asr():
                 start_time = time.monotonic()
@@ -132,8 +130,6 @@ class ASRProviderBase(ABC):
                     return None
             
             # 使用线程池执行器并行运行
-            parallel_start_time = time.monotonic()
-            
             with concurrent.futures.ThreadPoolExecutor(max_workers=2) as thread_executor:
                 asr_future = thread_executor.submit(run_asr)
                 
@@ -151,7 +147,7 @@ class ASRProviderBase(ABC):
             
             
             # 处理结果
-            raw_text, file_path = results.get("asr", ("", None))
+            raw_text, _ = results.get("asr", ("", None))
             speaker_name = results.get("voiceprint", None)
             
             # 记录识别结果

main/xiaozhi-server/core/providers/llm/AliBL/AliBL.py

@@ -1,8 +1,10 @@
 from config.logger import setup_logging
 from http import HTTPStatus
+import dashscope
 from dashscope import Application
 from core.providers.llm.base import LLMProviderBase
 from core.utils.util import check_model_key
+import time
 
 TAG = __name__
 logger = setup_logging()
@@ -15,6 +17,7 @@ class LLMProvider(LLMProviderBase):
         self.base_url = config.get("base_url")
         self.is_No_prompt = config.get("is_no_prompt")
         self.memory_id = config.get("ali_memory_id")
+        self.streaming_chunk_size = config.get("streaming_chunk_size", 3)  # 每次流式返回的字符数
         check_model_key("AliBLLLM", self.api_key)
 
     def response(self, session_id, dialogue):
@@ -32,6 +35,8 @@ class LLMProvider(LLMProviderBase):
                 "app_id": self.app_id,
                 "session_id": session_id,
                 "messages": dialogue,
+                # 开启SDK原生流式
+                "stream": True,
             }
             if self.memory_id != False:
                 # 百练memory需要prompt参数
@@ -42,25 +47,63 @@ class LLMProvider(LLMProviderBase):
                     f"【阿里百练API服务】处理后的prompt: {prompt}"
                 )
 
+            # 可选地设置自定义API基地址（若配置为兼容模式URL则忽略）
+            if self.base_url and ("/api/" in self.base_url):
+                dashscope.base_http_api_url = self.base_url
+
             responses = Application.call(**call_params)
-            if responses.status_code != HTTPStatus.OK:
-                logger.bind(tag=TAG).error(
-                    f"code={responses.status_code}, "
-                    f"message={responses.message}, "
-                    f"请参考文档：https://help.aliyun.com/zh/model-studio/developer-reference/error-code"
-                )
-                yield "【阿里百练API服务响应异常】"
-            else:
-                logger.bind(tag=TAG).debug(
-                    f"【阿里百练API服务】构造参数: {call_params}"
-                )
-                yield responses.output.text
+
+            # 流式处理（SDK在stream=True时返回可迭代对象；否则返回单次响应对象）
+            logger.bind(tag=TAG).debug(
+                f"【阿里百练API服务】构造参数: {dict(call_params, api_key='***')}"
+            )
+
+            last_text = ""
+            try:
+                for resp in responses:
+                    if resp.status_code != HTTPStatus.OK:
+                        logger.bind(tag=TAG).error(
+                            f"code={resp.status_code}, message={resp.message}, 请参考文档：https://help.aliyun.com/zh/model-studio/developer-reference/error-code"
+                        )
+                        continue
+                    current_text = getattr(getattr(resp, "output", None), "text", None)
+                    if current_text is None:
+                        continue
+                    # SDK流式为增量覆盖，计算差量输出
+                    if len(current_text) >= len(last_text):
+                        delta = current_text[len(last_text):]
+                    else:
+                        # 避免偶发回退
+                        delta = current_text
+                    if delta:
+                        yield delta
+                    last_text = current_text
+            except TypeError:
+                # 非流式回落（一次性返回）
+                if responses.status_code != HTTPStatus.OK:
+                    logger.bind(tag=TAG).error(
+                        f"code={responses.status_code}, message={responses.message}, 请参考文档：https://help.aliyun.com/zh/model-studio/developer-reference/error-code"
+                    )
+                    yield "【阿里百练API服务响应异常】"
+                else:
+                    full_text = getattr(getattr(responses, "output", None), "text", "")
+                    logger.bind(tag=TAG).info(
+                        f"【阿里百练API服务】完整响应长度: {len(full_text)}"
+                    )
+                    for i in range(0, len(full_text), self.streaming_chunk_size):
+                        chunk = full_text[i:i + self.streaming_chunk_size]
+                        if chunk:
+                            yield chunk
 
         except Exception as e:
             logger.bind(tag=TAG).error(f"【阿里百练API服务】响应异常: {e}")
             yield "【LLM服务响应异常】"
 
     def response_with_functions(self, session_id, dialogue, functions=None):
-        logger.bind(tag=TAG).error(
-            f"阿里百练暂未实现完整的工具调用（function call），建议使用其他意图识别"
+        # 阿里百练当前未支持原生的 function call。为保持兼容，这里回退到普通文本流式输出。
+        # 上层会按 (content, tool_calls) 的形式消费，这里始终返回 (token, None)
+        logger.bind(tag=TAG).warning(
+            "阿里百练未实现原生 function call，已回退为纯文本流式输出"
         )
+        for token in self.response(session_id, dialogue):
+            yield token, None

main/xiaozhi-server/core/providers/tools/server_mcp/mcp_executor.py

@@ -18,8 +18,8 @@ class ServerMCPExecutor(ToolExecutor):
         """初始化MCP管理器"""
         if not self._initialized:
             self.mcp_manager = ServerMCPManager(self.conn)
-            await self.mcp_manager.initialize_servers()
             self._initialized = True
+            await self.mcp_manager.initialize_servers()
 
     async def execute(
         self, conn, tool_name: str, arguments: Dict[str, Any]

main/xiaozhi-server/core/providers/tools/server_mcp/mcp_manager.py

@@ -68,6 +68,9 @@ class ServerMCPManager:
 
         # 输出当前支持的服务端MCP工具列表
         if hasattr(self.conn, "func_handler") and self.conn.func_handler:
+            # 刷新工具缓存以确保服务端MCP工具被正确加载
+            if hasattr(self.conn.func_handler, "tool_manager"):
+                self.conn.func_handler.tool_manager.refresh_tools()
             self.conn.func_handler.current_support_functions()
 
     def get_all_tools(self) -> List[Dict[str, Any]]:

main/xiaozhi-server/core/providers/tts/aliyun_stream.py

@@ -216,7 +216,6 @@ class TTSProvider(TTSProviderBase):
 
                 if message.sentence_type == SentenceType.FIRST:
                     self.conn.client_abort = False
-                    self.reset_flow_controller()
 
                 if self.conn.client_abort:
                     logger.bind(tag=TAG).info("收到打断信息，终止TTS文本处理线程")
@@ -479,3 +478,142 @@ class TTSProvider(TTSProviderBase):
         finally:
             self._monitor_task = None
 
+    def to_tts(self, text: str) -> list:
+        """非流式TTS处理，用于测试及保存音频文件的场景"""
+        try:
+            # 创建新的事件循环
+            loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(loop)
+
+            # 生成会话ID
+            session_id = uuid.uuid4().hex
+            # 存储音频数据
+            audio_data = []
+
+            async def _generate_audio():
+                # 刷新Token（如果需要）
+                if self._is_token_expired():
+                    self._refresh_token()
+
+                # 建立WebSocket连接
+                ws = await websockets.connect(
+                    self.ws_url,
+                    additional_headers={"X-NLS-Token": self.token},
+                    ping_interval=30,
+                    ping_timeout=10,
+                    close_timeout=10,
+                )
+                try:
+                    # 发送StartSynthesis请求
+                    start_message_id = str(uuid.uuid4().hex)
+                    start_request = {
+                        "header": {
+                            "message_id": start_message_id,
+                            "task_id": session_id,
+                            "namespace": "FlowingSpeechSynthesizer",
+                            "name": "StartSynthesis",
+                            "appkey": self.appkey,
+                        },
+                        "payload": {
+                            "voice": self.voice,
+                            "format": self.format,
+                            "sample_rate": self.sample_rate,
+                            "volume": self.volume,
+                            "speech_rate": self.speech_rate,
+                            "pitch_rate": self.pitch_rate,
+                            "enable_subtitle": True,
+                        },
+                    }
+                    await ws.send(json.dumps(start_request))
+
+                    # 等待SynthesisStarted响应
+                    synthesis_started = False
+                    while not synthesis_started:
+                        msg = await ws.recv()
+                        if isinstance(msg, str):
+                            data = json.loads(msg)
+                            header = data.get("header", {})
+                            if header.get("name") == "SynthesisStarted":
+                                synthesis_started = True
+                                logger.bind(tag=TAG).debug("TTS合成已启动")
+                            elif header.get("name") == "TaskFailed":
+                                error_info = data.get("payload", {}).get(
+                                    "error_info", {}
+                                )
+                                error_code = error_info.get("error_code")
+                                error_message = error_info.get(
+                                    "error_message", "未知错误"
+                                )
+                                raise Exception(
+                                    f"启动合成失败: {error_code} - {error_message}"
+                                )
+
+                    # 发送文本合成请求
+                    filtered_text = MarkdownCleaner.clean_markdown(text)
+                    run_message_id = str(uuid.uuid4().hex)
+                    run_request = {
+                        "header": {
+                            "message_id": run_message_id,
+                            "task_id": session_id,
+                            "namespace": "FlowingSpeechSynthesizer",
+                            "name": "RunSynthesis",
+                            "appkey": self.appkey,
+                        },
+                        "payload": {"text": filtered_text},
+                    }
+                    await ws.send(json.dumps(run_request))
+
+                    # 发送停止合成请求
+                    stop_message_id = str(uuid.uuid4().hex)
+                    stop_request = {
+                        "header": {
+                            "message_id": stop_message_id,
+                            "task_id": session_id,
+                            "namespace": "FlowingSpeechSynthesizer",
+                            "name": "StopSynthesis",
+                            "appkey": self.appkey,
+                        }
+                    }
+                    await ws.send(json.dumps(stop_request))
+
+                    # 接收音频数据
+                    synthesis_completed = False
+                    while not synthesis_completed:
+                        msg = await ws.recv()
+                        if isinstance(msg, (bytes, bytearray)):
+                            self.opus_encoder.encode_pcm_to_opus_stream(
+                                msg,
+                                end_of_stream=False,
+                                callback=lambda opus: audio_data.append(opus)
+                            )
+                        elif isinstance(msg, str):
+                            data = json.loads(msg)
+                            header = data.get("header", {})
+                            event_name = header.get("name")
+                            if event_name == "SynthesisCompleted":
+                                synthesis_completed = True
+                                logger.bind(tag=TAG).debug("TTS合成完成")
+                            elif event_name == "TaskFailed":
+                                error_info = data.get("payload", {}).get(
+                                    "error_info", {}
+                                )
+                                error_code = error_info.get("error_code")
+                                error_message = error_info.get(
+                                    "error_message", "未知错误"
+                                )
+                                raise Exception(
+                                    f"合成失败: {error_code} - {error_message}"
+                                )
+                finally:
+                    try:
+                        await ws.close()
+                    except:
+                        pass
+
+            loop.run_until_complete(_generate_audio())
+            loop.close()
+
+            return audio_data
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"生成音频数据失败: {str(e)}")
+            return []

main/xiaozhi-server/core/providers/tts/base.py

@@ -1,22 +1,22 @@
 import os
 import re
-import queue
+import time
 import uuid
+import queue
 import asyncio
 import threading
-from typing import Callable, Any
+import traceback
 from core.utils import p3
-import time
 from datetime import datetime
 from core.utils import textUtils
+from typing import Callable, Any
 from abc import ABC, abstractmethod
 from config.logger import setup_logging
-from core.utils.audio_flow_control import FlowControlConfig
-from core.utils.util import audio_bytes_to_data_stream, audio_to_data_stream
 from core.utils.tts import MarkdownCleaner
 from core.utils.output_counter import add_device_output
 from core.handle.reportHandle import enqueue_tts_report
 from core.handle.sendAudioHandle import sendAudioMessage
+from core.utils.util import audio_bytes_to_data_stream, audio_to_data_stream
 from core.providers.tts.dto.dto import (
     TTSMessageDTO,
     SentenceType,
@@ -24,8 +24,6 @@ from core.providers.tts.dto.dto import (
     InterfaceType,
 )
 
-import traceback
-
 TAG = __name__
 logger = setup_logging()
 
@@ -34,7 +32,6 @@ class TTSProviderBase(ABC):
     def __init__(self, config, delete_audio_file):
         self.interface_type = InterfaceType.NON_STREAM
         self.conn = None
-        self.tts_timeout = 10
         self.delete_audio_file = delete_audio_file
         self.audio_file_type = "wav"
         self.output_file = config.get("output_dir", "tmp/")
@@ -71,7 +68,6 @@ class TTSProviderBase(ABC):
         self.tts_stop_request = False
         self.processed_chars = 0
         self.is_first_sentence = True
-        self.flow_controller = FlowControlConfig.create_flow_controller()
 
     def generate_filename(self, extension=".wav"):
         return os.path.join(
@@ -147,6 +143,68 @@ class TTSProviderBase(ABC):
             except Exception as e:
                 logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
                 return None
+    
+    def to_tts(self, text):
+        text = MarkdownCleaner.clean_markdown(text)
+        max_repeat_time = 5
+        if self.delete_audio_file:
+            # 需要删除文件的直接转为音频数据
+            while max_repeat_time > 0:
+                try:
+                    audio_bytes = asyncio.run(self.text_to_speak(text, None))
+                    if audio_bytes:
+                        audio_datas = []
+                        audio_bytes_to_data_stream(
+                            audio_bytes,
+                            file_type=self.audio_file_type,
+                            is_opus=True,
+                            callback=lambda data: audio_datas.append(data)
+                        )
+                        return audio_datas
+                    else:
+                        max_repeat_time -= 1
+                except Exception as e:
+                    logger.bind(tag=TAG).warning(
+                        f"语音生成失败{5 - max_repeat_time + 1}次: {text}，错误: {e}"
+                    )
+                    max_repeat_time -= 1
+            if max_repeat_time > 0:
+                logger.bind(tag=TAG).info(
+                    f"语音生成成功: {text}，重试{5 - max_repeat_time}次"
+                )
+            else:
+                logger.bind(tag=TAG).error(
+                    f"语音生成失败: {text}，请检查网络或服务是否正常"
+                )
+            return None
+        else:
+            tmp_file = self.generate_filename()
+            try:
+                while not os.path.exists(tmp_file) and max_repeat_time > 0:
+                    try:
+                        asyncio.run(self.text_to_speak(text, tmp_file))
+                    except Exception as e:
+                        logger.bind(tag=TAG).warning(
+                            f"语音生成失败{5 - max_repeat_time + 1}次: {text}，错误: {e}"
+                        )
+                        # 未执行成功，删除文件
+                        if os.path.exists(tmp_file):
+                            os.remove(tmp_file)
+                        max_repeat_time -= 1
+
+                if max_repeat_time > 0:
+                    logger.bind(tag=TAG).info(
+                        f"语音生成成功: {text}:{tmp_file}，重试{5 - max_repeat_time}次"
+                    )
+                else:
+                    logger.bind(tag=TAG).error(
+                        f"语音生成失败: {text}，请检查网络或服务是否正常"
+                    )
+
+                return tmp_file
+            except Exception as e:
+                logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
+                return None
 
     @abstractmethod
     async def text_to_speak(self, text, output_file):
@@ -194,7 +252,6 @@ class TTSProviderBase(ABC):
 
     async def open_audio_channels(self, conn):
         self.conn = conn
-        self.tts_timeout = conn.config.get("tts_timeout", 10)
         # tts 消化线程
         self.tts_priority_thread = threading.Thread(
             target=self.tts_text_priority_thread, daemon=True
@@ -225,7 +282,6 @@ class TTSProviderBase(ABC):
                     self.tts_text_buff = []
                     self.is_first_sentence = True
                     self.tts_audio_first_sentence = True
-                    self.reset_flow_controller()
                 elif ContentType.TEXT == message.content_type:
                     self.tts_text_buff.append(message.content_detail)
                     segment_text = self._get_segment_text()
@@ -270,103 +326,43 @@ class TTSProviderBase(ABC):
 
                 if self.conn.client_abort:
                     logger.bind(tag=TAG).debug("收到打断信号，跳过当前音频数据")
-                    # 打断时丢弃未上报的音频数据
                     enqueue_text, enqueue_audio = None, []
                     continue
 
                 # 收到下一个文本开始或会话结束时进行上报
                 if sentence_type is not SentenceType.MIDDLE:
+                    # 重置音频流控状态（新句子开始或者结束）
+                    if hasattr(self.conn, 'audio_flow_control'):
+                        self.conn.audio_flow_control = {
+                            'last_send_time': 0,
+                            'packet_count': 0,
+                            'start_time': time.perf_counter()
+                        }
+                    
                     # 上报TTS数据
                     if enqueue_text is not None and enqueue_audio is not None:
                         enqueue_tts_report(self.conn, enqueue_text, enqueue_audio)
                     enqueue_audio = []
                     enqueue_text = text
 
-                # 计算音频数据的帧数
-                if isinstance(audio_datas, bytes):
-                    frame_count = 1  # 单个字节流作为一帧
+                # 收集上报音频数据
+                if isinstance(audio_datas, bytes) and enqueue_audio is not None:
                     enqueue_audio.append(audio_datas)
-                else:
-                    frame_count = 0
+
+                # 发送音频
+                future = asyncio.run_coroutine_threadsafe(
+                    sendAudioMessage(self.conn, sentence_type, audio_datas, text),
+                    self.conn.loop,
+                )
+                future.result()
 
                 # 记录输出和报告
                 if self.conn.max_output_size > 0 and text:
                     add_device_output(self.conn.headers.get("device-id"), len(text))
 
-                # 流控检查
-                if frame_count > 0:
-                    max_wait_time = FlowControlConfig.DEFAULT_MAX_WAIT_TIME
-                    wait_start_time = time.time()
-                    retry_interval = FlowControlConfig.DEFAULT_RETRY_INTERVAL
-
-                    while not self.flow_controller.can_send_frames(frame_count):
-                        # 检查是否超时或需要停止
-                        if (
-                            time.time() - wait_start_time > max_wait_time
-                            or self.conn.stop_event.is_set()
-                            or self.conn.client_abort
-                        ):
-                            logger.bind(tag=TAG).debug(
-                                "流控等待超时或收到停止信号，跳过音频发送"
-                            )
-                            break
-                        # 短暂等待后重试
-                        time.sleep(retry_interval)
-                    else:
-                        # 可以发送，记录发送的帧数
-                        self.flow_controller.record_sent_frames(frame_count)
-
-                        # 发送音频
-                        future = asyncio.run_coroutine_threadsafe(
-                            self._send_audio_with_flow_control(
-                                sentence_type, audio_datas, text
-                            ),
-                            self.conn.loop,
-                        )
-                        future.result()
-
-                        # 输出流控状态（调试用）
-                        # status = self.flow_controller.get_status()
-                        # logger.bind(tag=TAG).debug(
-                        #     f"流控状态: 缓冲区使用率={status['buffer_usage_percent']:.1f}%, "
-                        #     f"可用令牌={status['available_tokens']}..."
-                        #     f"发送帧数={status['sent_frames']}..."
-                        #     f"消费帧数={status['consumed_frames']}..."
-                        #     f"代播放帧数={status['sent_frames'] - status['consumed_frames']}..."
-                        # )
-                else:
-                    # 没有音频数据，直接发送
-                    future = asyncio.run_coroutine_threadsafe(
-                        self._send_audio_with_flow_control(
-                            sentence_type, audio_datas, text
-                        ),
-                        self.conn.loop,
-                    )
-                    future.result()
-
             except Exception as e:
                 logger.bind(tag=TAG).error(f"audio_play_priority_thread: {text} {e}")
 
-    async def _send_audio_with_flow_control(self, sentence_type, audio_datas, text):
-        """
-        带流控的音频发送方法 模拟设备消费音频帧的过程
-        实际应用中应该根据设备反馈来更新消费情况
-        """
-        await sendAudioMessage(self.conn, sentence_type, audio_datas, text)
-
-        # 模拟设备消费（实际应用中应该从设备获取反馈）防止音字不同步
-        if isinstance(audio_datas, bytes):
-            # 模拟设备播放延迟（60ms per frame）, 实际情况可以低一点（50ms），增加使用体验
-            await asyncio.sleep(0.06)
-            self.flow_controller.update_device_consumption(1)
-
-    # 在类中添加流控制器重置方法
-    def reset_flow_controller(self):
-        """重置流控制器状态，通常在新会话开始时调用"""
-        if hasattr(self, "flow_controller"):
-            self.flow_controller.reset()
-            logger.bind(tag=TAG).info("流控制器状态已重置")
-
     async def start_session(self, session_id):
         pass
 

main/xiaozhi-server/core/providers/tts/fishspeech.py

@@ -143,8 +143,8 @@ class TTSProvider(TTSProviderBase):
         data = {
             "text": text,
             "references": [
-                ServeReferenceAudio(audio=audio if audio else b"", text=text)
-                for text, audio in zip(ref_texts, byte_audios)
+                ServeReferenceAudio(audio=audio if audio else b"", text=ref_text)
+                for ref_text, audio in zip(ref_texts, byte_audios)
             ],
             "reference_id": self.reference_id,
             "normalize": self.normalize,

main/xiaozhi-server/core/providers/tts/huoshan_double_stream.py

@@ -213,7 +213,6 @@ class TTSProvider(TTSProviderBase):
 
                 if message.sentence_type == SentenceType.FIRST:
                     self.conn.client_abort = False
-                    self.reset_flow_controller()
 
                 if self.conn.client_abort:
                     try:
@@ -629,3 +628,104 @@ class TTSProvider(TTSProviderBase):
 
     def wav_to_opus_data_audio_raw_stream(self, raw_data_var, is_end=False, callback: Callable[[Any], Any]=None):
         return self.opus_encoder.encode_pcm_to_opus_stream(raw_data_var, is_end, callback=callback)
+
+    def to_tts(self, text: str) -> list:
+        """非流式生成音频数据，用于生成音频及测试场景
+        Args:
+            text: 要转换的文本
+        Returns:
+            list: 音频数据列表
+        """
+        try:
+            # 创建事件循环
+            loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(loop)
+
+            # 生成会话ID
+            session_id = uuid.uuid4().__str__().replace("-", "")
+
+            # 存储音频数据
+            audio_data = []
+
+            async def _generate_audio():
+                # 创建新的WebSocket连接
+                ws_header = {
+                    "X-Api-App-Key": self.appId,
+                    "X-Api-Access-Key": self.access_token,
+                    "X-Api-Resource-Id": self.resource_id,
+                    "X-Api-Connect-Id": uuid.uuid4(),
+                }
+                ws = await websockets.connect(
+                    self.ws_url, additional_headers=ws_header, max_size=1000000000
+                )
+
+                try:
+                    # 启动会话
+                    header = Header(
+                        message_type=FULL_CLIENT_REQUEST,
+                        message_type_specific_flags=MsgTypeFlagWithEvent,
+                        serial_method=JSON,
+                    ).as_bytes()
+                    optional = Optional(
+                        event=EVENT_StartSession, sessionId=session_id
+                    ).as_bytes()
+                    payload = self.get_payload_bytes(
+                        event=EVENT_StartSession, speaker=self.voice
+                    )
+                    await self.send_event(ws, header, optional, payload)
+
+                    # 发送文本
+                    header = Header(
+                        message_type=FULL_CLIENT_REQUEST,
+                        message_type_specific_flags=MsgTypeFlagWithEvent,
+                        serial_method=JSON,
+                    ).as_bytes()
+                    optional = Optional(
+                        event=EVENT_TaskRequest, sessionId=session_id
+                    ).as_bytes()
+                    payload = self.get_payload_bytes(
+                        event=EVENT_TaskRequest, text=text, speaker=self.voice
+                    )
+                    await self.send_event(ws, header, optional, payload)
+
+                    # 发送结束会话请求
+                    header = Header(
+                        message_type=FULL_CLIENT_REQUEST,
+                        message_type_specific_flags=MsgTypeFlagWithEvent,
+                        serial_method=JSON,
+                    ).as_bytes()
+                    optional = Optional(
+                        event=EVENT_FinishSession, sessionId=session_id
+                    ).as_bytes()
+                    payload = str.encode("{}")
+                    await self.send_event(ws, header, optional, payload)
+
+                    # 接收音频数据
+                    while True:
+                        msg = await ws.recv()
+                        res = self.parser_response(msg)
+
+                        if (
+                            res.optional.event == EVENT_TTSResponse
+                            and res.header.message_type == AUDIO_ONLY_RESPONSE
+                        ):
+                            self.wav_to_opus_data_audio_raw_stream(res.payload, callback=lambda opus_frame: audio_data.append(opus_frame))
+                        elif res.optional.event == EVENT_SessionFinished:
+                            break
+
+                finally:
+                    # 清理资源
+                    try:
+                        await ws.close()
+                    except:
+                        pass
+
+            # 运行异步任务
+            loop.run_until_complete(_generate_audio())
+            loop.close()
+
+            return audio_data
+
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"生成音频数据失败: {str(e)}")
+            return []

main/xiaozhi-server/core/providers/tts/index_stream.py

@@ -1,8 +1,10 @@
 import os
+import time
 import queue
-import asyncio
-import traceback
 import aiohttp
+import asyncio
+import requests
+import traceback
 from config.logger import setup_logging
 from core.utils.tts import MarkdownCleaner
 from core.providers.tts.base import TTSProviderBase
@@ -45,7 +47,6 @@ class TTSProvider(TTSProviderBase):
                     self.processed_chars = 0
                     self.tts_text_buff = []
                     self.before_stop_play_files.clear()
-                    self.reset_flow_controller()
                 elif ContentType.TEXT == message.content_type:
                     self.tts_text_buff.append(message.content_detail)
                     segment_text = self._get_segment_text()
@@ -178,3 +179,57 @@ class TTSProvider(TTSProviderBase):
         await super().close()
         if hasattr(self, "opus_encoder"):
             self.opus_encoder.close()
+
+    def to_tts(self, text: str) -> list:
+        """非流式TTS处理，用于测试及保存音频文件的场景
+        Args:
+            text: 要转换的文本
+        Returns:
+            list: 返回opus编码后的音频数据列表
+        """
+        start_time = time.time()
+        text = MarkdownCleaner.clean_markdown(text)
+
+        payload = {"text": text, "character": self.character}
+
+        try:
+            with requests.post(self.api_url, json=payload, timeout=5) as response:
+                if response.status_code != 200:
+                    logger.bind(tag=TAG).error(
+                        f"TTS请求失败: {response.status_code}, {response.text}"
+                    )
+                    return []
+
+                logger.info(f"TTS请求成功: {text}, 耗时: {time.time() - start_time}秒")
+
+                # 使用opus编码器处理PCM数据
+                opus_datas = []
+                pcm_data = response.content
+
+                # 计算每帧的字节数
+                frame_bytes = int(
+                    self.opus_encoder.sample_rate
+                    * self.opus_encoder.channels
+                    * self.opus_encoder.frame_size_ms
+                    / 1000
+                    * 2
+                )
+
+                # 分帧处理PCM数据
+                for i in range(0, len(pcm_data), frame_bytes):
+                    frame = pcm_data[i : i + frame_bytes]
+                    if len(frame) < frame_bytes:
+                        # 最后一帧可能不足，用0填充
+                        frame = frame + b"\x00" * (frame_bytes - len(frame))
+
+                    self.opus_encoder.encode_pcm_to_opus_stream(
+                        frame,
+                        end_of_stream=(i + frame_bytes >= len(pcm_data)),
+                        callback=lambda opus: opus_datas.append(opus)
+                    )
+
+                return opus_datas
+
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"TTS请求异常: {e}")
+            return []

main/xiaozhi-server/core/providers/tts/linkerai.py

@@ -1,8 +1,10 @@
 import os
+import time
 import queue
-import asyncio
-import traceback
 import aiohttp
+import asyncio
+import requests
+import traceback
 from config.logger import setup_logging
 from core.utils.tts import MarkdownCleaner
 from core.providers.tts.base import TTSProviderBase
@@ -42,7 +44,6 @@ class TTSProvider(TTSProviderBase):
                     self.processed_chars = 0
                     self.tts_text_buff = []
                     self.before_stop_play_files.clear()
-                    self.reset_flow_controller()
                 elif ContentType.TEXT == message.content_type:
                     self.tts_text_buff.append(message.content_detail)
                     segment_text = self._get_segment_text()
@@ -110,10 +111,6 @@ class TTSProvider(TTSProviderBase):
         finally:
             return None
 
-    ###################################################################################
-    # linkerai单流式TTS重写父类的方法--结束
-    ###################################################################################
-
     async def text_to_speak(self, text, is_last):
         """流式处理TTS音频，每句只推送一次音频列表"""
         await self._tts_request(text, is_last)
@@ -200,3 +197,71 @@ class TTSProvider(TTSProviderBase):
         except Exception as e:
             logger.bind(tag=TAG).error(f"TTS请求异常: {e}")
             self.tts_audio_queue.put((SentenceType.LAST, [], None))
+
+    def to_tts(self, text: str) -> list:
+        """非流式TTS处理，用于测试及保存音频文件的场景
+        Args:
+            text: 要转换的文本
+        Returns:
+            list: 返回opus编码后的音频数据列表
+        """
+        start_time = time.time()
+        text = MarkdownCleaner.clean_markdown(text)
+
+        params = {
+            "tts_text": text,
+            "spk_id": self.voice,
+            "frame_duration": 60,
+            "stream": False,
+            "target_sr": 16000,
+            "audio_format": self.audio_format,
+            "instruct_text": "请生成一段自然流畅的语音",
+        }
+        headers = {
+            "Authorization": f"Bearer {self.access_token}",
+            "Content-Type": "application/json",
+        }
+
+        try:
+            with requests.get(
+                self.api_url, params=params, headers=headers, timeout=5
+            ) as response:
+                if response.status_code != 200:
+                    logger.bind(tag=TAG).error(
+                        f"TTS请求失败: {response.status_code}, {response.text}"
+                    )
+                    return []
+
+                logger.info(f"TTS请求成功: {text}, 耗时: {time.time() - start_time}秒")
+
+                # 使用opus编码器处理PCM数据
+                opus_datas = []
+                pcm_data = response.content
+
+                # 计算每帧的字节数
+                frame_bytes = int(
+                    self.opus_encoder.sample_rate
+                    * self.opus_encoder.channels
+                    * self.opus_encoder.frame_size_ms
+                    / 1000
+                    * 2
+                )
+
+                # 分帧处理PCM数据
+                for i in range(0, len(pcm_data), frame_bytes):
+                    frame = pcm_data[i : i + frame_bytes]
+                    if len(frame) < frame_bytes:
+                        # 最后一帧可能不足，用0填充
+                        frame = frame + b"\x00" * (frame_bytes - len(frame))
+
+                    self.opus_encoder.encode_pcm_to_opus_stream(
+                        frame,
+                        end_of_stream=(i + frame_bytes >= len(pcm_data)),
+                        callback=lambda opus: opus_datas.append(opus)
+                    )
+
+                return opus_datas
+
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"TTS请求异常: {e}")
+            return []

main/xiaozhi-server/core/providers/tts/paddle_speech.py

@@ -1,13 +1,15 @@
-import asyncio
-import json
-import base64
-import aiohttp
-import numpy as np
 import io
 import wave
+import json
+import base64
+import asyncio
 import websockets
-from core.providers.tts.base import TTSProviderBase
+import numpy as np
+from datetime import datetime
 from config.logger import setup_logging
+from core.providers.tts.base import TTSProviderBase
+
+
 
 TAG = __name__
 logger = setup_logging()
@@ -18,11 +20,12 @@ class TTSProvider(TTSProviderBase):
         super().__init__(config, delete_audio_file)
         self.url = config.get("url", "ws://192.168.1.10:8092/paddlespeech/tts/streaming")
         self.protocol = config.get("protocol", "websocket")
+        
         if config.get("private_voice"):
             self.spk_id = int(config.get("private_voice"))
         else:
-            self.spk_id = int(config.get("spk_id", "0"))  
-        
+            self.spk_id = int(config.get("spk_id", "0"))
+            
         sample_rate = config.get("sample_rate", 24000)
         self.sample_rate = float(sample_rate) if sample_rate else 24000
         
@@ -32,7 +35,21 @@ class TTSProvider(TTSProviderBase):
         volume = config.get("volume", 1.0)
         self.volume = float(volume) if volume else 1.0
         
-        self.save_path = config.get("save_path", "./streaming_tts.wav")
+        self.delete_audio_file = config.get("delete_audio", True)
+        if not self.delete_audio_file:
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            save_path = config.get("save_path")
+            if save_path:
+                if not save_path.endswith('.wav'):
+                    save_path = f"{save_path}_{timestamp}.wav"
+                else:
+                    other_path = save_path[:-4]
+                    save_path = f"{other_path}_{timestamp}.wav"
+                self.save_path = save_path
+            else:
+                self.save_path = f"./streaming_tts_{timestamp}.wav"
+        else:
+            self.save_path = None
 
     async def pcm_to_wav(self, pcm_data: bytes, sample_rate: int = 24000, num_channels: int = 1,
                          bits_per_sample: int = 16) -> bytes:
@@ -58,43 +75,9 @@ class TTSProvider(TTSProviderBase):
     async def text_to_speak(self, text, output_file):
         if self.protocol == "websocket":
             return await self.text_streaming(text, output_file)
-        elif self.protocol == "http":
-            return await self.text(text, output_file)
         else:
             raise ValueError("Unsupported protocol. Please use 'websocket' or 'http'.")
 
-    async def text(self, text, output_file):
-        request_json = {
-            "text": text,
-            "spk_id": self.spk_id,
-            "speed": self.speed,
-            "volume": self.volume,
-            "sample_rate": self.sample_rate,
-            "save_path": self.save_path
-        }
-
-        try:
-            async with aiohttp.ClientSession() as session:
-                async with session.post(self.url, json=request_json) as resp:
-                    if resp.status == 200:
-                        resp_json = await resp.json()
-                        if resp_json.get("success"):
-                            data = resp_json["result"]
-                            audio_bytes = base64.b64decode(data["audio"])
-                            if output_file:
-                                with open(output_file, "wb") as file_to_save:
-                                    file_to_save.write(audio_bytes)
-                            else:
-                                return audio_bytes
-                        else:
-                            raise Exception(
-                                f"Error: {resp_json.get('message', 'Unknown error')} while processing text: {text}")
-                    else:
-                        raise Exception(
-                            f"HTTP Error: {resp.status} - {await resp.text()} while processing text: {text}")
-        except Exception as e:
-            raise Exception(f"Error during TTS HTTP request: {e} while processing text: {text}")
-
     async def text_streaming(self, text, output_file):
         try:
             # 使用 websockets 异步连接到 WebSocket 服务器
@@ -151,6 +134,12 @@ class TTSProvider(TTSProviderBase):
                 # 接收结束响应避免服务抛出异常
                 await ws.recv()
 
+                # 根据配置决定是否保存文件
+                if not self.delete_audio_file and self.save_path:
+                    with open(self.save_path, "wb") as f:
+                        f.write(wav_data)
+                    logger.bind(tag=TAG).info(f"音频文件已保存到: {self.save_path}")
+                
                 # 返回或保存音频数据
                 if output_file:
                     with open(output_file, "wb") as file_to_save:
@@ -159,4 +148,4 @@ class TTSProvider(TTSProviderBase):
                     return wav_data
 
         except Exception as e:
-            raise Exception(f"Error during TTS WebSocket request: {e} while processing text: {text}")
+            raise Exception(f"Error during TTS WebSocket request: {e} while processing text: {text}")

main/xiaozhi-server/core/utils/audio_flow_control.py

@@ -1,186 +0,0 @@
-"""
-音频流控模块
-包含令牌桶算法和音频流控制器的实现
-"""
-
-import asyncio
-import time
-import threading
-from collections import deque
-from typing import Optional, Dict, Any
-
-
-class TokenBucket:
-    """令牌桶实现，用于限流控制"""
-
-    def __init__(self, capacity: int, refill_rate: float, initial_tokens: Optional[int] = None):
-        """
-        初始化令牌桶
-
-        Args:
-            capacity: 桶容量（最大令牌数）
-            refill_rate: 令牌补充速率（每秒补充的令牌数）
-            initial_tokens: 初始令牌数，默认为桶容量
-        """
-        self.capacity = capacity
-        self.refill_rate = refill_rate
-        self.tokens = initial_tokens if initial_tokens is not None else capacity
-        self.last_refill_time = time.time()
-        self.lock = threading.Lock()
-
-    def get_tokens(self, requested_tokens: int = 1) -> bool:
-        """
-        获取指定数量的令牌
-
-        Args:
-            requested_tokens: 请求的令牌数量
-
-        Returns:
-            bool: 是否成功获取到令牌
-        """
-        with self.lock:
-            self._refill_tokens()
-
-            if self.tokens >= requested_tokens:
-                self.tokens -= requested_tokens
-                return True
-            else:
-                return False
-
-    def get_available_tokens(self) -> int:
-        """获取当前可用令牌数"""
-        with self.lock:
-            self._refill_tokens()
-            return int(self.tokens)
-
-    def _refill_tokens(self):
-        """内部方法：补充令牌"""
-        current_time = time.time()
-        time_passed = current_time - self.last_refill_time
-        tokens_to_add = time_passed * self.refill_rate
-
-        self.tokens = min(self.capacity, self.tokens + tokens_to_add)
-        self.last_refill_time = current_time
-
-
-class AudioFlowController:
-    """音频流控制器，基于令牌桶算法控制音频数据发送"""
-
-    def __init__(self, max_device_buffer: int = 3000, refill_rate: float = 20):
-        """
-        初始化音频流控制器
-
-        Args:
-            max_device_buffer: 设备端最大缓冲区大小（Opus帧数）
-            refill_rate: 令牌补充速率（每秒允许发送的帧数）
-        """
-        self.max_device_buffer = max_device_buffer
-        self.token_bucket = TokenBucket(
-            capacity=max_device_buffer,
-            refill_rate=refill_rate,
-            initial_tokens=max_device_buffer // 2  # 初始令牌为容量的一半
-        )
-        self.sent_frames_count = 0  # 已发送帧数计数
-        self.device_consumed_frames = 0  # 设备端已消费帧数
-        self.pending_queue = deque()  # 等待发送的数据队列
-        self._lock = threading.Lock()
-
-    def can_send_frames(self, frame_count: int) -> bool:
-        """
-        检查是否可以发送指定数量的帧
-
-        Args:
-            frame_count: 要发送的帧数
-
-        Returns:
-            bool: 是否可以发送
-        """
-        with self._lock:
-            # 检查设备端缓冲区是否会溢出
-            estimated_device_buffer = self.sent_frames_count - self.device_consumed_frames
-            if estimated_device_buffer + frame_count > self.max_device_buffer:
-                return False
-
-            # 检查令牌桶是否有足够令牌
-            return self.token_bucket.get_tokens(frame_count)
-
-    def update_device_consumption(self, consumed_frames: int):
-        """
-        更新设备端消费的帧数
-
-        Args:
-            consumed_frames: 设备端消费的帧数
-        """
-        with self._lock:
-            self.device_consumed_frames += consumed_frames
-
-    def record_sent_frames(self, frame_count: int):
-        """
-        记录已发送的帧数
-
-        Args:
-            frame_count: 发送的帧数
-        """
-        with self._lock:
-            self.sent_frames_count += frame_count
-
-    def get_status(self) -> Dict[str, Any]:
-        """获取流控状态信息"""
-        with self._lock:
-            estimated_buffer = self.sent_frames_count - self.device_consumed_frames
-            return {
-                "sent_frames": self.sent_frames_count,
-                "consumed_frames": self.device_consumed_frames,
-                "estimated_device_buffer": estimated_buffer,
-                "available_tokens": self.token_bucket.get_available_tokens(),
-                "pending_queue_size": len(self.pending_queue),
-                "buffer_usage_percent": (estimated_buffer / self.max_device_buffer) * 100
-            }
-
-    def reset(self):
-        """重置流控状态"""
-        with self._lock:
-            self.sent_frames_count = 0
-            self.device_consumed_frames = 0
-            self.pending_queue.clear()
-            # 重新初始化令牌桶
-            self.token_bucket = TokenBucket(
-                capacity=self.max_device_buffer,
-                refill_rate=self.token_bucket.refill_rate,
-                initial_tokens=self.max_device_buffer // 2
-            )
-
-
-# 流控配置常量
-class FlowControlConfig:
-    """流控配置常量"""
-    # Opus 编码参数
-    OPUS_FRAME_DURATION_MS = 60  # Opus帧时长（毫秒）
-    OPUS_FRAMES_PER_SECOND = 1000 / OPUS_FRAME_DURATION_MS  # 每秒帧数
-
-    # 默认流控参数
-    DEFAULT_MAX_DEVICE_BUFFER = 40  # 设备端最大缓冲帧数
-    DEFAULT_REFILL_RATE = OPUS_FRAMES_PER_SECOND  # 默认令牌补充速率（帧/秒）
-    DEFAULT_MAX_WAIT_TIME = 5.0  # 流控最大等待时间（秒）
-    DEFAULT_RETRY_INTERVAL = 0.06  # 流控重试间隔（秒）
-
-    # 预缓冲参数
-    PRE_BUFFER_FRAMES = 3  # 预缓冲帧数
-
-    @classmethod
-    def create_flow_controller(cls, max_buffer: Optional[int] = None,
-                               refill_rate: Optional[float] = None) -> AudioFlowController:
-        """
-        创建流控制器的工厂方法
-
-        Args:
-            max_buffer: 最大缓冲区大小，使用默认值如果为None
-            refill_rate: 令牌补充速率，使用默认值如果为None
-
-        Returns:
-            AudioFlowController: 配置好的流控制器实例
-        """
-        return AudioFlowController(
-            max_device_buffer=max_buffer or cls.DEFAULT_MAX_DEVICE_BUFFER,
-            refill_rate=refill_rate or cls.DEFAULT_REFILL_RATE
-        )

main/xiaozhi-server/core/utils/opus_encoder_utils.py

@@ -6,10 +6,9 @@ Opus编码工具类
 import logging
 import traceback
 import numpy as np
-from typing import Optional, Callable, Any
 from opuslib_next import Encoder
 from opuslib_next import constants
-
+from typing import Optional, Callable, Any
 
 class OpusEncoderUtils:
     """PCM到Opus的编码器"""
@@ -130,4 +129,4 @@ class OpusEncoderUtils:
     def close(self):
         """关闭编码器并释放资源"""
         # opuslib没有明确的关闭方法，Python的垃圾回收会处理
-        pass
+        pass

main/xiaozhi-server/core/utils/p3.py

@@ -1,12 +1,15 @@
-import io
 import struct
-from typing import Callable, Any
 
+def decode_opus_from_file(input_file):
+    """
+    从p3文件中解码 Opus 数据，并返回一个 Opus 数据包的列表以及总时长。
+    """
+    opus_datas = []
+    total_frames = 0
+    sample_rate = 16000  # 文件采样率
+    frame_duration_ms = 60  # 帧时长
+    frame_size = int(sample_rate * frame_duration_ms / 1000)
 
-def decode_opus_from_file_stream(input_file, callback: Callable[[Any], Any]):
-    """
-    从p3文件中解码 Opus 数据，由 callback 处理 Opus 数据包。
-    """
     with open(input_file, 'rb') as f:
         while True:
             # 读取头部（4字节）：[1字节类型，1字节保留，2字节长度]
@@ -22,13 +25,23 @@ def decode_opus_from_file_stream(input_file, callback: Callable[[Any], Any]):
             if len(opus_data) != data_len:
                 raise ValueError(f"Data length({len(opus_data)}) mismatch({data_len}) in the file.")
 
-            callback(opus_data)
+            opus_datas.append(opus_data)
+            total_frames += 1
 
+    # 计算总时长
+    total_duration = (total_frames * frame_duration_ms) / 1000.0
+    return opus_datas, total_duration
 
-def decode_opus_from_bytes_stream(input_bytes, callback: Callable[[Any], Any]):
+def decode_opus_from_bytes(input_bytes):
     """
-    从p3二进制数据中解码 Opus 数据，由 callback 处理 Opus 数据包。
+    从p3二进制数据中解码 Opus 数据，并返回一个 Opus 数据包的列表以及总时长。
     """
+    import io
+    opus_datas = []
+    total_frames = 0
+    sample_rate = 16000  # 文件采样率
+    frame_duration_ms = 60  # 帧时长
+    frame_size = int(sample_rate * frame_duration_ms / 1000)
 
     f = io.BytesIO(input_bytes)
     while True:
@@ -39,4 +52,8 @@ def decode_opus_from_bytes_stream(input_bytes, callback: Callable[[Any], Any]):
         opus_data = f.read(data_len)
         if len(opus_data) != data_len:
             raise ValueError(f"Data length({len(opus_data)}) mismatch({data_len}) in the bytes.")
-        callback(opus_data)
+        opus_datas.append(opus_data)
+        total_frames += 1
+
+    total_duration = (total_frames * frame_duration_ms) / 1000.0
+    return opus_datas, total_duration

main/xiaozhi-server/core/utils/util.py

@@ -1,16 +1,17 @@
-import json
-import socket
-import subprocess
 import re
 import os
-from io import BytesIO
-from typing import Callable, Any
-from core.utils import p3
-import numpy as np
-import requests
-import opuslib_next
-from pydub import AudioSegment
+import json
 import copy
+import wave
+import socket
+import requests
+import subprocess
+import numpy as np
+import opuslib_next
+from io import BytesIO
+from core.utils import p3
+from pydub import AudioSegment
+from typing import Callable, Any
 
 TAG = __name__
 emoji_map = {
@@ -228,6 +229,56 @@ def audio_to_data_stream(audio_file_path, is_opus=True, callback: Callable[[Any]
     raw_data = audio.raw_data
     pcm_to_data_stream(raw_data, is_opus, callback)
 
+def audio_to_data(audio_file_path: str, is_opus: bool = True) -> list[bytes]:
+    """
+    将音频文件转换为Opus/PCM编码的帧列表
+    Args:
+        audio_file_path: 音频文件路径
+        is_opus: 是否进行Opus编码
+    """
+    # 获取文件后缀名
+    file_type = os.path.splitext(audio_file_path)[1]
+    if file_type:
+        file_type = file_type.lstrip(".")
+    # 读取音频文件，-nostdin 参数：不要从标准输入读取数据，否则FFmpeg会阻塞
+    audio = AudioSegment.from_file(
+        audio_file_path, format=file_type, parameters=["-nostdin"]
+    )
+
+    # 转换为单声道/16kHz采样率/16位小端编码（确保与编码器匹配）
+    audio = audio.set_channels(1).set_frame_rate(16000).set_sample_width(2)
+
+    # 获取原始PCM数据（16位小端）
+    raw_data = audio.raw_data
+
+    # 初始化Opus编码器
+    encoder = opuslib_next.Encoder(16000, 1, opuslib_next.APPLICATION_AUDIO)
+
+    # 编码参数
+    frame_duration = 60  # 60ms per frame
+    frame_size = int(16000 * frame_duration / 1000)  # 960 samples/frame
+
+    datas = []
+    # 按帧处理所有音频数据（包括最后一帧可能补零）
+    for i in range(0, len(raw_data), frame_size * 2):  # 16bit=2bytes/sample
+        # 获取当前帧的二进制数据
+        chunk = raw_data[i : i + frame_size * 2]
+
+        # 如果最后一帧不足，补零
+        if len(chunk) < frame_size * 2:
+            chunk += b"\x00" * (frame_size * 2 - len(chunk))
+
+        if is_opus:
+            # 转换为numpy数组处理
+            np_frame = np.frombuffer(chunk, dtype=np.int16)
+            # 编码Opus数据
+            frame_data = encoder.encode(np_frame.tobytes(), frame_size)
+        else:
+            frame_data = chunk if isinstance(chunk, bytes) else bytes(chunk)
+
+        datas.append(frame_data)
+
+    return datas
 
 def audio_bytes_to_data_stream(audio_bytes, file_type, is_opus, callback: Callable[[Any], Any]) -> None:
     """
@@ -273,6 +324,31 @@ def pcm_to_data_stream(raw_data, is_opus=True, callback: Callable[[Any], Any] =
             frame_data = chunk if isinstance(chunk, bytes) else bytes(chunk)
             callback(frame_data)
 
+def opus_datas_to_wav_bytes(opus_datas, sample_rate=16000, channels=1):
+    """
+    将opus帧列表解码为wav字节流
+    """
+    decoder = opuslib_next.Decoder(sample_rate, channels)
+    pcm_datas = []
+
+    frame_duration = 60  # ms
+    frame_size = int(sample_rate * frame_duration / 1000)  # 960
+
+    for opus_frame in opus_datas:
+        # 解码为PCM（返回bytes，2字节/采样点）
+        pcm = decoder.decode(opus_frame, frame_size)
+        pcm_datas.append(pcm)
+
+    pcm_bytes = b"".join(pcm_datas)
+
+    # 写入wav字节流
+    wav_buffer = BytesIO()
+    with wave.open(wav_buffer, "wb") as wf:
+        wf.setnchannels(channels)
+        wf.setsampwidth(2)  # 16bit
+        wf.setframerate(sample_rate)
+        wf.writeframes(pcm_bytes)
+    return wav_buffer.getvalue()
 
 def check_vad_update(before_config, new_config):
     if (

main/xiaozhi-server/core/utils/wakeup_word.py

@@ -0,0 +1,140 @@
+import os
+import re
+import yaml
+import time
+import hashlib
+import portalocker
+from typing import Dict
+
+
+class FileLock:
+    def __init__(self, file, timeout=5):
+        self.file = file
+        self.timeout = timeout
+        self.start_time = None
+
+    def __enter__(self):
+        self.start_time = time.time()
+        while True:
+            try:
+                portalocker.lock(self.file, portalocker.LOCK_EX | portalocker.LOCK_NB)
+                return self.file
+            except portalocker.LockException:
+                if time.time() - self.start_time > self.timeout:
+                    raise TimeoutError("获取文件锁超时")
+                time.sleep(0.1)
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        portalocker.unlock(self.file)
+
+
+class WakeupWordsConfig:
+    def __init__(self):
+        self.config_file = "data/.wakeup_words.yaml"
+        self.assets_dir = "config/assets/wakeup_words"
+        self._ensure_directories()
+        self._config_cache = None
+        self._last_load_time = 0
+        self._cache_ttl = 1  # 缓存有效期（秒）
+        self._lock_timeout = 5  # 文件锁超时时间（秒）
+
+    def _ensure_directories(self):
+        """确保必要的目录存在"""
+        os.makedirs(os.path.dirname(self.config_file), exist_ok=True)
+        os.makedirs(self.assets_dir, exist_ok=True)
+
+    def _load_config(self) -> Dict:
+        """加载配置文件，使用缓存机制"""
+        current_time = time.time()
+
+        # 如果缓存有效，直接返回缓存
+        if (
+            self._config_cache is not None
+            and current_time - self._last_load_time < self._cache_ttl
+        ):
+            return self._config_cache
+
+        try:
+            with open(self.config_file, "a+") as f:
+                with FileLock(f, timeout=self._lock_timeout):
+                    f.seek(0)
+                    content = f.read()
+                    config = yaml.safe_load(content) if content else {}
+                    self._config_cache = config
+                    self._last_load_time = current_time
+                    return config
+        except (TimeoutError, IOError) as e:
+            print(f"加载配置文件失败: {e}")
+            return {}
+        except Exception as e:
+            print(f"加载配置文件时发生未知错误: {e}")
+            return {}
+
+    def _save_config(self, config: Dict):
+        """保存配置到文件，使用文件锁保护"""
+        try:
+            with open(self.config_file, "w") as f:
+                with FileLock(f, timeout=self._lock_timeout):
+                    yaml.dump(config, f, allow_unicode=True)
+                    self._config_cache = config
+                    self._last_load_time = time.time()
+        except (TimeoutError, IOError) as e:
+            print(f"保存配置文件失败: {e}")
+            raise
+        except Exception as e:
+            print(f"保存配置文件时发生未知错误: {e}")
+            raise
+
+    def get_wakeup_response(self, voice: str) -> Dict:
+        voice = hashlib.md5(voice.encode()).hexdigest()
+        """获取唤醒词回复配置"""
+        config = self._load_config()
+
+        if not config or voice not in config:
+            return None
+
+        # 检查文件大小
+        file_path = config[voice]["file_path"]
+        if not os.path.exists(file_path) or os.stat(file_path).st_size < (15 * 1024):
+            return None
+
+        return config[voice]
+
+    def update_wakeup_response(self, voice: str, file_path: str, text: str):
+        """更新唤醒词回复配置"""
+        try:
+            # 过滤表情符号
+            filtered_text = re.sub(r'[\U0001F600-\U0001F64F\U0001F900-\U0001F9FF]', '', text)
+            
+            config = self._load_config()
+            voice_hash = hashlib.md5(voice.encode()).hexdigest()
+            config[voice_hash] = {
+                "voice": voice,
+                "file_path": file_path,
+                "time": time.time(),
+                "text": filtered_text,
+            }
+            self._save_config(config)
+        except Exception as e:
+            print(f"更新唤醒词回复配置失败: {e}")
+            raise
+
+    def generate_file_path(self, voice: str) -> str:
+        """生成音频文件路径，使用voice的哈希值作为文件名"""
+        try:
+            # 生成voice的哈希值
+            voice_hash = hashlib.md5(voice.encode()).hexdigest()
+            file_path = os.path.join(self.assets_dir, f"{voice_hash}.wav")
+
+            # 如果文件已存在，先删除
+            if os.path.exists(file_path):
+                try:
+                    os.remove(file_path)
+                except Exception as e:
+                    print(f"删除已存在的音频文件失败: {e}")
+                    raise
+
+            return file_path
+        except Exception as e:
+            print(f"生成音频文件路径失败: {e}")
+            raise